[Brotkrumen]

Imagine you are designing a software system. One use case requires you to model a fact as best as you can, so you write a fact-class for it. Your software system runs fine, but you notice the class doesn't accurately model the fact in some other use cases. You now know the class is insufficiently accurate. Do you still trust the result of your software?

Replace software with language, class with concept we use to refer to facts and you have to claim that any further thought into the design of your class is wasted, because you hold that to be the case for concepts, language and reality.

[asfasgasg]

If all philosophers are talking about is how underspecified words are, that is fine, but there's little value in it. Language is not designed. So a word is underspecified. Nobody can fix it.

But I was under the impression that philosophers thought themselves more than mere dictionary debaters. Whenever I hear about this problem, my impression is that at least non-zero of them think there is a problem with the actual concept of knowledge, not merely the word.